Contrastive marking in French dialogue: Why and how.
Monique Vion, Annie Colas
To cite this version:
Monique Vion, Annie Colas. Contrastive marking in French dialogue: Why and how.. Journal
of Psycholinguistic Research, Springer Verlag, 1995, 24 (5), pp.313-331.
HAL Id: hal-00134163
https://hal.archives-ouvertes.fr/hal-00134163
Submitted on 1 Mar 2007
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Contrastive Marking1 in French Dialogue: Why and How
Monique Vion and Annie Colas2
This experiment studies French-speaking adults ' preferencesfor prosodic marking (focal accent)
or nzoiphosyntactic marking (clefling) to express information contrasts in dialogue. Our goal is to
determine what syntactic and conceptual factors might contribute to these preferences, by
examining for the former, the grammatical function of the item bearing the contrastive mark
(subject vs complement) and, for the latter, the size of the class to which the contrasted item
belongs (two members vs more than two members,). The subjects' responses on a forced-choice
judgment task showed that when only one device was usedfor contrast, subjects clearly preferred
cleflingfor grammatical subjects andfocal accentfor complements. When both devices were used
(prosodic and morphosyntactic), contrasted subjects were preferred over contrasted
complements. Response times were longer when the contrasted item belonged to a two-member
class. These results demonstrate that subjects' judgments of the suitability of linguistic devices for
expressing information contrasts in French are more highly affected by syntactic factors than
conceptual ones.
INTRODUCTION
During dialogue, speakers formulate their statemnts in accordance with what
they know (or think they know) about their addressee (experience, skills, current
preoccupation, etc.). As the conversation progresses, the speaker compiles and
updates a mental model of the conversational partner (Clark & Clark, 1977;
Steedman & Johnson-Laird, 1980). This model incorporates a number of
presumptions about the knowledge and beliefs the addressee is currently
enteiaining. In an attempt to establish common ground; the speaker makes
assumptions about the information available to the addressee for interpreting
what is being said. He/she also assumes that there is somè information the
listener does not know (Clark & Carson, 1981; Clark & Marshall, 1981; Clark & 1laviland, 1977; Haviland & Clark, 1974). As the dialogue devélops, the speaker
may be led to believe that the information currently in the listener's mind is not
applicable to the conveisation. Engaged in a collaborative referring process
(Clark & Wilkes-Gibbs, 1986; Clark & Schaefer, 1987, 1989), the speaker may
then attempt to change the partner's beliefs by introducing an information
contrast into the subsequent statements.
1
2
CREPCO URA 182, CNRS, Université de Provence, Aix-en-Provence, France.
Address all correspondence to Monique Vion, CREPCO URA 182, CNRS, Université de
Provence, 29 Avenue Robert-Schuman, F-13621 Aix-en-Provence Cedex I , France.
A contrast is a form of implicit negation of what the audience believes.
Formally, two oppositions are made: a syntagmatic opposition and a paradigmatic opposition (Touati, 1987). At the syntagmatic level, a contrast is used to
focus on a particular element, i.e., to make it stand out as the most important
element in the information being conveyed. At the paradigmatic level, a contrast
opposes the important element to all other items the addressee might consider in
its place. As Chafe (1974, 1976) observed, the speaker who marks a contrast
simultaneously does three things: he/she (i) assumes that the listener has at
his/her disposal a restricted set of items which might contribute to understanding
the statements being made, (ii) selects a particular item among them, and (iii)
states that it is the right one. Below, dialogue 1 illustrates this process.
(1)
A: Il parait que les Martin ont eu un accident de voiture?
'It looks like the Martins had a car accident?'
B: Oui, ils n'ont pas vu le stop et ils ont heurté un camion.
'Yes, they didn't see the stop sign and they hit a truck.'
A: Ha! Ces hommes, toujours pressés.
'Oh, men! Always in a hurry.'
B: C'est Madame Martin qui conduisait.
'It's Mrs. Martin who was driving.'
Approaching the problem from a linguistic standpoint, a number of studies
have proposed functional descriptions for languages spoken in Europe (Borkin,
1984; Bromser, 1984; Chafe, 1974, 1976; Halliday, 1967, 1970; Prince, 1978;
Sornicola, 1988). These studies have pointed out two types of devices used to
make a contrast: clefting and focal accent.
Clefting has been well documented, at both the syntactic (Knowles, 1986;
Moreau, 1976) and the semantic-pragmatic levels. A cleft sentence uses two
clauses to express what could have been said in one. In French, clefting is
achieved by means of the expression "c'est ... qu" ("it's ...that/which/who"), which
has two morphological variants, "c'est ... qui" and "c'est ... que," depending on
the grammatical function of the clefted constituent. "C'est ... qui" is used when
the clefted constituent is the subject of the sentence (regardless of its semantic
role). The form "c'est . . que' ' is used when the clefted constituent is a
complement (example 2):
(2)
(a) Pierre a invité Jean.
(2) 'Peter invited John.'
(b) C'est Pierre qui a invité Jean.
'It's Peter who invited Jean.'
(c) C 'est Jean que Pierre a invité.
'It's John that Peter invited.'
Clefting can be used with different constituents of a clause (example 3) or even
with an entire proposition (example 4).
(3)
(a) Pierre in 'a raconté hier des souvenirs de son enfance.
'Péter told me stories yesterday about his childhood.'
(b) C'est Pierre qui m'a raconté hier des souvenirs de son enfance.
'It's Peter who told me stories yesterday about his childhood.'
(c) C'est des souvenirs de son enfance que Pierre ni 'a raconté
hier.
'It's stories about his childhood that Peter told me yesterday.'
(d) C'est à moi que Pierre a raconté hier des souvenirs de son
enfance.
'It's to me that Peter told stories yesterday about his childhood.'
(e) C'est hier que Pierre m'a raconté des souvenirs de son enfance.
'It's yesterday that Peter told me stories about his childhood.'
(4)
C 'est parce qu'it' aimait trop les femmes que Pierre ne s'est pas
marie.
'It's because he loved women too much that Peter never married.'
.
From a pragmatic standpoint, the clefted constituent is the focus of the
assertion. Clefting may correlate with the contrastive function, as in example 5.
(5)
C'est la fatigue qui gagne Jean.
'It's fatigue that affects John.'
But as we shall see below, clefting and/or focal accent can be used to obtain a
variety of contrastive effects.
Linguistic studies dealing with prosodic phenomena3 such as stress are
more recent and probably less familiar to psycholinguists. In French4, there are
several kinds of stress, one of which is focal ccent, of interest to us here (Di
Cristo, 1995). Focal accent is used to rank the elements of an utterance, and
therefore plays a pragmatic role. It highlights one of the lexical items in the
utterance, as in example 6 below5.
(6)
(a) l'AMI de Nini a volé des habits
3
Prosody is the general terni used to refer to phenomena such as stress, tone, rhythm. and
intonation.
4
Rossi (1985) distinguishes among three types of stress in French. Internal accent generates
intonation. It is a property of the morpheme. External accent is so called because it is not a
property of the morpheme and generally does not coincide with internal accent. It is also called
focal accent, and fulfills a pragmatic function. Melodic ictus is manifested by a melodic
prominence in sequences of more than five syllables without internal accent or intonation
morphemes. It appears to be governed by a rhythmic constraint and is located at the point where
an external accent could potentially occur.
5
Di Cristo, personal communication. The stressed word is shown in capital letters.
'Nini's FRIEND stole some clothes'
(b) l'ami de NINI a volé des habits
'NINI'S friend stole some clothes'
(c) l'ami de Nini A VOLE des habits
'Nini's friend STOLE some clothes'
(d) l'ami de Nini a volé des HABITS
'Nini's friend stole some CLOTHES'
The function of focal accent in French is to "rhematize" one of the
elements of the utterance in cases where there are syntactic constraints which
prohibit the occurrence of the conclusive intonation morpheme6, the device
ordinarily used to fulfill the rhematic function. Focal accent does not necessarily
play a contrastive role, but it can be used to this end and is often mistakenly
taken for emphatic stress. The two accents should not be confused since focal
accent has no intrinsic expressive value (although expressive content may be
added to it) (Rossi, 1985). One characteristic of focal accent in French is a rapid
change in pitch: "As a general rule an item focussed for contrast is characterized
by a global rising-falling pitch pattern. While the fall is always associated with the
rightmost full syllable of the item in contrast, the timing of the rising movement
lending prominence depends on the objective or expressive character of the
focus' (Di Cristo, 1995).
Our empirical study examines the case where the same constituent can
be contrasted by clefling, focal accent, or both, as in example 7.
(7)
(a) PIERRE écrit des articles pour un journal.
'PETER writes articles for a newspaper.'
(b) C'est Pierre qui écrit des articles pour un journal.
'It's Peter who writes articles for a newspaper.'
(c) C'est PIERRE qui écrit des articles pour un journal.
'It's PETER who writes articles for a newspaper.'
But this is not the only case, as linguistic analyses have pointed out7. There are
distinctions that can be rendered by using cleft sentences but cannot be easily
expressed by focal accent alone. Consider example 8.
(8)
(a) C'est Jean que je n 'ai pas vu.
'It's John that I didn't see.'
6
The conclusive intonation morpheme, which also marks the syntactic organization of the
utterance, "is is realized by a melodic fall or a static tone in the lower or infra-lower pitch range; it
is accompanied by an intensity glissando of about - 10 dB and a duration increase of
approximately 100% relative to the mean of unstressed vowels" (Rossi,1985, p. 141).
7
We would like to thank an anonymous reviewer for his or her constructive criticism and
examples, which led us to add the section that follows.
(b) Ce n'est pas Jean que j'ai vu.
'It isn't John that I saw.'
(c) C 'est n'est pas Jean que je n'ai pas vu.
'It isn't John that I didn't see.'
Example 8a can be used contrastively to correct the impression that I didn't see
someone other than John (in the universe of discourse), resulting in the inference
that I saw everyone else; example 8b, that I saw John, resulting in the inference
that I saw someone other than John; and example 8c, that I didn't see John,
resulting in the inference that I didn't see someone other than John. The focally
stressed example 9 can be used only to cover the situations expressed in
examples 8a and 8b, and is, in fact, ambiguous in that regard.
(9)
Je n’ai pas vu JEAN.
'I didn't see JOHN.'
Moreover, when the clefted constituent is grammatically complex, it is
possible that only part of the constituent is contrasted (example 10).
(10)
(a) C'est le chat de PIERRE qui s'est perdu.
'It's PETER's cat that got lost.'
(b) C'est le CHAT de Pierre qui s'est perdu.
'It Peter's CAT that got lost.'
It is also possible, albeit unusual in ordinary discourse, that a cleft
construction is used while the contrast concerns a constituent other than the
clefted one. In this case the other constituent is highlighted by a focal accent
(example 11).
(11)
C 'est la fatigue qui gagne JEAN.
'It's fatigue that affects JOHN.'
Although "fatigue" is the clefted constituent, "John" is the focused item.
Sentences like this can be used to correct an interlocutor's impression that
someone other than , John (in the universe of discourse) is the one whom fatigue
affects.
In studies with a psycholinguistic approach to the problem, some empirical
data are available on cleft sentence processing in English and French (Amy &
Vion, 1986; Bronckart, 1979; Hornby, 1971, 1974; Vion & Amy, 1984) and on the
function of cleft constructions in dialogue (Hupet & Tilmant, 1986; Tilmant &
Hupet, 1990). There are also some empirical data on the processing of stress in
English and German for isolated words (Cutler, 1976; Cutler & Foss, 1977,
Welsh, 1980) and sentences (Bock & Mazella, 1983; Hornby & Hass, 1970;
Maratsos, 1973; Needham, 1990; Pechman, 1.984; Terken & Nooteboom, 1987).
Among the above studies, those using judgments about the suitability of
contrastive utterances with or without clefting (Hupet & Tilmant, 1986) have
shown that cleft constructions are preferred to the uncleft form with the same
meaning. However, the utterances judged by the subjects have been given in
written form. The authors themselves agree that in the written medium,
potentially contrastive uncleft utterances are stripped of the intonation cues they
would have in the oral medium. This medium also prevents the study of the
extent to which exclusively prosodic marking via focal accent is selected. The
preferences previously observed with written dialogues may no longer be
obtained for contrastive utterances in oral conversation, where morphosyntactic
and/or prosodic marking can be proposed. The following experiment was
designed to include prosodic marking in the study of preferences for different
ways of expressing a contrast.
In addition to the question of whether subjects' preferences will change
with a greater range of contrasting devices is the question of the reasons for
those preferences. Several possible explanations might be considered. First,
they may depend on cognitive constraints. A contrast is defined in reference to
the knowledge the listener is assumed to have. To make a contrast, the speaker
must assume that there are other possibilities available to the listener. When
there are two items in this set of possibilities (often a pair of
Contrastive Marking in Dialogue 319
antonyms), the erroneous item the listener has in mind makes the other item in
the pair highly accessible, as in example 1. When the set contains more than two
items, each individual candidate is less accessible. In the latter case, clefting
may be less suitable than focal accent for marking the contrast, since, as Terken
and Nooteboom (1987) contend, focal accent increases the availability of the
piece of information being accentuated.
Second, subjects' preferences may also stem from linguistic constraints
which limit the application of prosodic rules. According to Hupet and Tuniant
(1986), it is generally prohibited to put an accent on the first item in a French
sentence (usually the subject-agent). The authors consider this to be the reason
why agent clefting is more frequent than patient clefting in both induced and
spontaneous speech production (Hupet, 1988). Placing stress for contrast on
words occurring later in the sentence appears to be more likely.
Finally, subjects' preferences may be rooted in their knowledge of the
pragmatic and social context of the utterance (Bronckart, 1985). The social
setting in which a dialogue is initiated, its goal, and the status and respective
ages of the addressees-all of these factors impose constraints on how subjects
assess the verbalizations of a speaker. Dialogue creates a typical interaction
framework where temporary "offenses" are always possible. This setting forces
the partners to engage in a process8 which allows them to anticipate and repair
such offenses. In this framework, the implicit negation of the addressee's
assumption, generated by the contrast, should not be perceived as potentially
jeopardizing the social images of the partners. Thus, subjects' preferences for
certain types of marking may originate from the fact that they are taking these
8
This process includes whatever a person does to avoid losing face, including ill fi-ont of
him/herself (Goffman, 1973, 1974, 1981).
constraints into account. (Even failing to emphasize a listener's misconception
via a contrast is a way of preventing him/her from losing face.)
This experiment attempts to neutralize the effects of the interaction
context. It focuses on the linguistic and/or cognitive factors that might determine
subjects' preferences for prosodic and/or morphosyntactic marking of contrasts.
METHOD
Task
Subjects were asked to listen to a series of short dialogues and make a
judgment concerning the ending remark in each. Two formulations of the last
remark were proposed, and the subjects had to select the one they considered to
be the most suitable for that particular exchange. To choose between the first
and the second formulation (forced choice), the subject pressed the ' ' 1 ' ' key or
the "2" ' key on the computer used in the experiment. The variables recorded
were the key pressed and the time taken to press it.
A title was displayed on the screen before each dialogue so that the
subject could activate a knowledge structure to guide dialogue comprehension
(Bransford & Johnson, 1972; Dooling & Mullet, 1973). The title described the
setting of the dialogue (see the list of dialogue topics in the Appendix).
Factors and Design
The effects of two factors were studied using a 2 X 2 factorial design. The
first factor (the syntactic factor) was the grammatical function of the item bearing
the contrast. In our case, the item was either the subject or the complement of
the utterance to be judged. The second factor (the conceptual factor) was the
size of the set of items the speaker might assume the listener was considering
for the focused item. In our case, the set contained either two members, or more
than two members. By combining these two factors, we could study four types of
contrasts (see Table I for an example of each).
The characteristics of the replies to be judged were obtained by combining
two variables, each with two possible formulations for marking the contrast:
clefting (C) vs no clefting (nC) and focal accent (A) vs no focal accent (nA). The
resulting four formulations (CA, nCA, CnA, and nCnA) were paired, making six
possible pairs of replies: (CA, nCA), (CA, CnA), (CA, nCnA), (nCA, CnA), (nCA,
nCnA), and (CnA, nCnA).
A given subject processed all four contrasts and all six pairs of replies. For
a given type of contrast, it took six dialogues to present the reply pairs (a
dialogue was assigned to only one of the possible pairs). The replies in a given
pair were formulated using the appropriate lexical items for that dialogue. Each
subject judged 24 dialogues.
In order to neutralize potential semantic-pragmatic effects linked to
dialogue topic, each of the six pairs was associated with each dialogue, and six
groups of subjects were established (Table II).
To neutralize potential effects of reply presentation order within each pair,
one order was used for half of the subjects and the reverse order for the other
half.
Subjects
Sixty native French-speaking students (male and female) from the
University of Provence participated in the experiment. They were tested
individually in a quiet room.
Table I. Types of Contrasts*
Grammatical
function
Subject
Complement
Set size
Two
More
A renter and an apartment manager
At the race track
- And what about the plumbing - So the race is over!
problems?
- It was close all the way.
- They've been solved at last.
- Obviously, with winners like that!
- The owner finally took care of it.
Fourth remark:
- It’s the renter who took care of it.
- The renter took care of it.
- It’s the renter who took care of it.'
- The renter took care of it.
Fourth remark:
- It’s Golden Arrow who won the race.
- Golden Arrow won the race.
- It’s Golden Arrow who won the race.
- Golden'Arrow won the race.
About getting a job
- You must be glad you got a job!
- Yes, I was hired as a school teacher.
- The Board of Education doesn't pay
very well!
Football: After drawing names
- So who are you playing with after all?
- We were pretty scared!
- It'll be easy to win the Brive team.
Fourth remark:
- It’s in a private school that I'm hired.
- I'm hired in a private school.
- It’s in a private school that I'm hired.
- I'm hired in a private school.
Fourth remark:
- It’s against Saint-Etienne that we'll be
playing.
- We'll be playing against Saint-Etienne.
- It’s against Saint-Etienne that t we'll be
playing.
- We'll be playing against Saint-Etienne.
*Morphosyntactic marks of clefting are underlined. Items bearing a focal accent are in italics.
Material and Procedure
Twenty-seven dialogues were generated, 3 to familiarize the subjects with
the task and 24 to test the effects of the experimental factors. The characters
chosen for the dialogues were familiar to each other (friends, neighbors, co-
workers), so that the conversations would neither start nor end by confirmatory
exchanges (Table I presents a sample dialogue for each cell in the experimental
design). The practice dialogues were generated in the same manner as the test
dialogues (only the structure of the replies differed; see example in the
Appendix).
Each dialogue consisted of four speaking turns which alternated between a male
and a female speaker. The female made the first and third remarks. After the third
remark, two formulations of the male speaker's reply were heard in succession. The
same speakers were recorded for all dialogues9.
Table II. Experimental Design
Subject two
1
2
3
4
5
Subject more
6
7
8
9
10
11
Type of contrasts
Complement two
Dialogue topics
12
13
14
Complement more
15
16
17
18
19
20
21
22
23
24
3
2
1
6
5
4
4
3
2
1
6
5
5
4
3
2
1
6
6
5
4
3
2
1
1
6
5
4
3
2
2
1
6
5
4
3
3
2
1
6
5
4
4
3
2
1
6
5
5
4
3
2
1
6
6
5
4
3
2
1
Reply pairs
1
6
5
4
3
2
2
1
6
5
4
3
3
2
1
6
5
4
4
3
2
1
6
5
5
4
3
2
1
6
6
5
4
3
2
1
1
6
5
4
3
2
2
1
6
5
4
3
3
2
1
6
5
4
4
3
2
1
6
5
5
4
3
2
1
6
6
5
4
3
2
1
1
6
5
4
3
2
2
1
6
5
4
3
Each test dialogue was recorded four times, once with each of the four
formulations of the male speaker's final remark. The tempo of the conversation
was maintained. Then for each dialogue, two listeners were asked to state which
recording of the first three remarks sounded the most "natural." This recording
was selected and saved in a computer file using the SoundEdit feature of the
MacRecorder system (for Macintosh SE30). The four formulations of the fourth
remark for each dialogue were also entered and saved in the computer. The
"Hypertalk"10 program used to run the experiment generated the pairs and
assigned them to the dialogue beginnings.
A constant time interval separated the displayed title from the first
speaking turn in each dialogue. The time between the end of the third remark
and the two formulations of the fourth was also constant. More time was allotted
for the latter interval (than that allotted between the first three remarks) to make
sure that the first formulation in the pair would not blend with the preceding
remarks and to allow subjects enough time to judge the two replies. A preliminary
survey with 24 subjects showed that a 1.5-sec interval was considered optimal
9
Our thanks to G6rard Amy and Elisabeth Dussauze for agreeing to act as speakers in the
dialogues.
10
We extend our gratitude to Jean-Claude Robert for his constant assistance during the writing of
the program.
for understanding the first three remarks and choosing one of the two replies
(Massip & Vilm, 1987).
The practice dialogues were given first, and then the 24 test dialogues
were presented. These two phases of the experiment were run in immediate
succession, without any mention of the changeover.
The subjects' instructions were as follows.
You are going to hear 27 short dialogues. Each dialogue is preceded by a title indicating
the topic. Each dialogue is composed of four remarks. You will hear the first three
remarks, a pause, and then the fourth remark. Two different formulations of the fourth
remark will be given, one after the other. When you have heard both, your task will be to
state which one seems to you to be the most appropriate to end the dialogue. To choose
the first, press key 1; to choose the second, press key 2. Respond immediately with the
first answer that comes to your mind. The time you take to decide will be recorded.
At the end of the experiment, the subjects were given the opportunity to
comment upon what choices they had made and why they had made them.
Predictions
We predicted that the responses obtained for the experiment as a whole
could be ranked by frequency of occurrence. The number of choices was
expected to be the lowest for formulations with no marks of contrast (nCIIA) and
the highest for ones with two mark9 (CA). One-mark formulations (CnA and nCA)
would fall somewhere in between.
Remember that the experiment was designed to study the effects of
syntactic and conceptual constraints on judgments of the suitability or
unsuitability of morphosyntactic and/or prosodic marks. If the choice between
clefting and focal accent depends on the grammatical function of the focused
item, then subjects should prefer clefting for subject contrasts and focal accent
for complement contrasts. If the choice between clefting and focal accent
depends on the size of the reference set, we can predict that clefting will be
chosen more often in cases where there is one alternative to the incorrect item in
the listener's mind (won vs lost, true vs false), whereas we can predict, in the
light of the scarce empirical data currently available, that focal accent will be
considered more suitable if not.
Although it has been suggested that response time is not a reliable
measure of processing when central cognitive activity is involved, we chose to
examine these data nonetheless. The task proposed here (lid indeed involve a
cognitive activity of this type, since it required subjects to make a metalinguistic
judgment about the last reply in a dialogue of which they started to build, a global
mental representation based on the title read and the remarks already heard.
However, because the instructions asked them to respond as spontaneously as
possible, and because the memory traces of the to-be-judged remarks were
transient (the previous remarks could not be reexamined as they might be in the
written mediun!), there was indeed some pressure on the subjects to answer
without delay. The fact that task execution was necessarily rapid suggested that,
although no specific predictions could be made, some conclusions might be
drawn from the response time data.
RESULTS
For each cell in our experimental design, a given subject could choose a
given type of reply (defined by combining the syntactic factor and the conceptual
factor) up to three times. And, for each of these cases, a dialogue could be
preferred up to 30 times with a given reply. Thus, each formulation could be
chosen at most 180 times per cell (60 subjects X 3 responses = 6 dialogues X 30
responses), making a total of 720 choices for the whole experiment.
The number of choices observed for each formulation relative to the total
number of times that choice could have been made was consistent with our
predictions. Formulations with no contrastive marks were chosen the least often
(nCnA = 43.6%). These were followed by formulations with clefting (CnA 46.6%)
and focal accent (nCA = 51.3%). Those combining clefting and focal accent were
chosen the most often (CA = 58%). Essentially the same tendency was noted for
the proportion of subjects who preferred a given formulation the majority of the
time. Table III indicates how many subjects chose the same formulation at least 8
times of 1 2 (X3 = 31.40, p < .001). Note that formulations with focal accent only
(nCA) were chosen more often than formulations with clefting only (CnA).
To test for the effect of the syntactic and conceptual factors on the choices
made, two analyses of variance with a 2 X 2 factorial design were conducted for
each type of formulation, first with the subjects and then with the dialogues as the
random factor. These analyses showed that for formulations containing a
contrastive mark of any kind (CA, CnA, and nCA), only the grammatical function
of the contrasted item triggered significant effects (Table IV).
Table IlI. Number of Subjects Making the Same Choice the Majority of the Time,
by Formulation (N = 60)
Criterion
Chosen at least 6 times of 12
Chosen at least 8 times of 12
nCnA
27
10
CnA
26
9
nCA
37
15
CA
41
23
These results are consistent with the predictions that utterances with
clefting only (CnA) would be chosen more frequently when the contrasted item
was the grammatical subject and that those with focal accent only (nCA) would
be chosen more frequently when the contrasted item was a complement. The
effect was not as statistically clear-cut for utterances bearing both contrastive
marks (CA). These utterances combined a mark which was more prone to be
accepted for grammatical subjects (clefting), with a mark which was more prone
to be accepted for complements (focal accent). The observed preference in the
CA case was comparable to the one observed for cleft-only utterances, namely,
a preference for sentences where the focused item was the grammatical subject.
In these cases, morphosyntactic marking via clefting prevailed over prosodic
marking.
Table V presents the total time taken to respond for each formulation.
Note in general that responding took approximately 1 sec.
The only possible way to examine response time was to analyze the data
with the dialogues as the random factor, since for a given cell (obtained by
combining the syntactic and conceptual factors), the number of responses
obtained for each dialogue varied (but was nonzero), whereas a given subject
may have never chosen a given formulation. The results given below should be
understood solely as preliminary indications-as food for thought for designing
further research. Analyses of each formulation expressing a contrast were
conducted for the mean and mdian response times (the statistical data given
below pertain to the medians).
Table IV. Mean Number of Choices, by Formulation
CA
NCA
CnA
By subject, DV (0,3)
Grammatical
function
Subj.
Compl.
F(l,59)
1.9
1.6
6.929
1.2
1.8
38.070
1.7
1.2
30.727
p
.0108
.00001
.00001
By dialogue, DV (0.30)
Grammatical
function
Subj.
Compl.
F(l,20)
p
18.8
15.9
4.256
.0523
12.4
18.3
27.913
.00001
16.8
11.1
26.499
.00001
Table V. Response Time, by Formulation
Response time
(msec)
Mean
Median
CnA
nCA
CA
1092
900
1175
758
1141
817
The only significant effects obtained for the means concerned the
dualmarked formulations (CA). The same effects Were found for the median
response times. Contrary to the findings obtained above for selection frequency,
an effect was found here for every factor. Responding was faster when the
focused item was a complement [F(l,20) 4.398, p .0489]. This means that CA
formulations with a focused complement were chosen less often, but more
rapidly. The response times were also shorter when the reference set contained
more than two members [F(1,20) 9.188, p.0066]; this effect was more
pronounced than the preceding one. Dual marking seems to have been chosen
more rapidly when the candidate items were less accessible. A tendential
interaction was also obtained between these two factors [F(l,20) = 3.4O9, p =
.0797], indicating a greater difference between the response times for
grammatical subjects (longer times) and complements (shorter times) when the
reference set was small than when it was large (Fig. 1). An analysis of the main
effects showed that the shorter response times (observed for complements
regardless of the size of the reference set and for grammatical subjects when the
reference set had more than two members) did not differ significantly from each
other. Only the response times observed for grammatical subjects with a twoitem reference set differed from those observed in the other three cases
[syntactic factor effect at the "two" ' level, F(1,20) 7.776, p .011; conceptual factor
effect at the "subject" level F(1,20) = 11.895, p = .003].
Although a similar effect was not observed for mean response times, a
significant effect of the conceptual factor for the clefting-only formulation (CnA)
was obtained on median response times. As with CA, clefting alone was chosen
more rapidly when the reference set contained more than two items [F(1,20)
9.624, p - .0056].
Fig.1. Interaction between the syntactic factor and the conceptual factors.
DISCUSSION
The objective of the present study was twofold. First, unlike previous
studies, our aim was to put subjects in experimental conditions which resembled
the real-world circumstances. in which contrastive utterances are encountered,
namely, in a conversational setting. Second, and more importantly, we hoped to
gain insight into how linguistic and cognitive factors affect subjects' preferences
for a given type of contrastive marking in speech.
Regarding the first objective, the experimental device was designed so
that we could collect judgments of oral dialogues. This allowed us to show that
subjects preferred utterances with prosodic marking. In the type of dialogue
chosen (dialogues between peers), utterances containing a contrastive focal
accent (alone or combined with clefting) were preferred over utterances with
clefting alone. To obtain a more complete view of subjects' preferences for
prosodic, syntactic, or morphosyntactic marking, the study could be extended in
two directions using the same experimental device. One could use contrastive
utterances in which several types of marking are often combined, as in negative
utterances such as "Non, Madame Martin conduisait' ' (No, Mrs. Martin was
driving) vs "Non, c'est Madame Martin qui conduisait" (No, it was Mrs. Martin who
was driving) vs "Non, MADAME Martin conduisait" (No, MRS. Martin was
driving). And one could use a wider variety of dialogues whre contrastive
utterances are found. This would provide a better picture of how metalinguistic
judgments vary as a function of the pragmatic aspects of the situation.
The second objective-the main goal of this study-was pursued in order to
provide evidence of the importance of two factors: the grammatical function of
the contrasted item and the size of the set to which it belongs. The experiment
demonstrated the importance of the syntactic factor in the judgments made. As
predicted, to mark a contrasted item, subjects preferred cleft constructions when
the item was the grammatical subject and focal accent when it was a
complement. Relative to the findings of previous studies, this result presents
three new points. (1) It shows that the same constraints govern judgment and
production (as studied so far). In both cases, subjects prefer clefting over focal
accent for making contrasts when the focused item is the grammatical subject.
(2) It shows, as Hupet and Tilmant (1986) suspected, that subjects prefer to use
focal accent to contrast complements. (3) Finally and above all, it shows that
syntactic constraints are
Contrastive Marking in Dialogue 327
(CnA) was obtained on median response times. As with CA, clefling alone was
chosen more rapidly when the reference set contained more than two items
[F(1,20) = 9.624, p .0056].
DISCUSSION
The objective of the present study was twofold. First, unlike previous
studies, our aim was to put subjects in experimental conditions which resembled
the real-world circumstances in which contrastive utterances are encountered,
namely, in a conversational setting. Second, and more importantly, we hoped to
gain insight into how linguistic and cognitive factors affect subjects' preferences
for a given type of contrastive marking in speech.
Regarding the first objective, the experimental device was designed so
that we could collect judgments of oral dialogues. This allowed us to show that
subjects preferred utterances with prosodic marking. In the type of dialogue
chosen (dialogues between peers), utterances containing a contrastive focal
accent (alone or combined with clefting) were preferred over utterances with
clefting alone. To obtain a more complete view of subjects' preferences for
prosodic, syntactic, or morphosyntactic marking, the study could be extended in
two directions using the same experimental device. One could use contrastive
utterances in which several types of marking are often combined, as in negative
utterances such as "Non, Madame Martin conduisait' ' (No, Mrs. Martin was
driving) vs "Non, c'est Madame Martin qui conduisait" (No, it was Mrs. Martin who
was driving) vs "Non, MADAME Martin conduisait' ' (No, MRS. Martin was
driving). And one could use a wider variety of dialogues whre contrastive
utterances are found. This would provide a better picture of how metalinguistic
judgments vary as a function of the pragmatic aspects of the situation.
The second objective-the main goal of this study-was pursued in order to
provide evidence of the importance of two factors: the grammatical function of
the contrasted item and the size of the set to which it belongs. The experiment
demonstrated the importance of the syntactic factor in the judgments made. As
predicted, to mark a contrasted item, subjects preferred cleft constructions when
the item was the grammatical subject and focal accent when it was a
complement. Relative to the findings of previous studies, this result presents
three new points. (1) It shows that the same constraints govern judgment and
production (as studied so far). In both cases, subjects prefer clefting over focal
accent for making contrasts when the focused item is the grammatical subject.
(2) It shows, as Hupet and Tilmant (1986) suspected, that subjects prefer to use
focal accent to contrast complements. (3) Finally and above all, it shows that
syntactic constraints are not the only ones, since subjects also consider it
acceptable to use focal accent on a grammatical subject placed at the beginning
of a sentence, and clefting for a complement. This experiment did not provide
evidence for an effect of the conceptual factor, as it was defined here, on the
nature of the judgments made. Once again, the experimental device allowed us
to measure the time taken by subjects to respond. The analyses of the response
times were indicative of how cognitive constraints might act in interaction with
linguistic constraints. In this experiment where subjects usually chose the
formulation with both clefting and focal accent, the response times were such
that although dual-marked utterances were chosen more frequently when the
focused item was the grammatical subject, responding was faster when it was a
complement. This finding can be interpreted in terms of the allocation of
attentional resources, as suggested in Hornby's (1973) study. Hornby showed
that when listeners had to state whether a drawing correctly depicted a sentence
they heard, but were not given enough time to explore the entire drawing, they
tended to focus their attention on the part of the utterance which was not the
object of presupposition (generally the conveyor of new information). Similarly,
the task proposed in the present experiment required the listeners to process the
dialogues within a limited time frame. We can assume that they prepared
themselves for judging the part of the utterance that most likely conveyed new
information, namely, the complement. This attentional preparation may have
allowed them to make quick choices, regardless of the size of the reference set,
when the contrast involved a complement. On the other hand, when the
contrasted item was a grammatical subject, their attention would have had to be
shifted. In this case, they may have become sensitive to the size of the reference
set, responding faster for sets with a larger number of items. The utility of dual
marking of contrasts may come more quickly to the mind of the listener when the
candidates are less accessible. This explanation opens up a road for future
research, although other experiments must be designed to explore this question
in greater depth.
APPENDIX
Dialogue Topics
Subject Two
After the conference
Repairing a car
The day after the maneuvers
Unfinished business
A renter and an apartment manager
The ups-and-downs of construction
Subject More
After a rugby game
About a car race
At the race tracks
A round of French bowls
At the National Assembly.
Before a tennis match
Complement Two
Preparing for vacation
Complement More
Leaving for the service
About getting a job
Buying clothes
Chess tournament
About travelling
Last-minute reservations
Football: after drawing names
About contracts
At a café
Writing award
Signing up for youth activities
Sample Practice Dialogue
At the Office
•Oh, Mr. Binet, you look upset! .
•Yes, my daughter is having health problems again. .
•She already had bronchitis last year ... .
•Two years ago.
•That was two years ago.
REFERENCES
Amy, G., & Vion, M. (1986). Les indices de traitement des phrases clivées. In C.
Bonnet, J.-P. Caverni, & J.-P. Codol (Eds.), Jugement et Langage. Bulletin
de Psychologie, 39(375), 377-386.
Bock, J. K., & Mazella, J. R. (1983). Intonational marking of Given and New
information: Some consequences for comprehension. Memory and
Cognition, 11, 64-76.
Borkin, A. (1984). Problems inform andfunction. Norwood, NJ: Ablex.
Bransford, J. D., & Johnson, M. K. (1972). Contextual prerequisites for
understanding: Some investigations of comprehension and recall. Journal
of Verbal Learning and Verbal Behavior, 11, 717-726.
Bromser, B. (1984). Towards a functional description of cleft constructions.
Lingua. 62, 325-348.
Bronckart, J. P. (1979). L'élaboration des opérations langagières: un exemple à
propos des structures à fonctions casuelles. Cahiers de l’Institut
Linguistique de Louvain, 5, 139-157.
Bronckart, J. P. (1985). Pour un modèle de production du discours. In J. P.
Bronckart (Ed.), Le fonctionnement des discours (pp. 3-58). Neuchâtel:
Delachaux et Niestlé.
Chafe, W. L. (1974). Language and consciousness. Language, 50, 111-133.
Chafe, W. L. (1976). Giveness, contrastiveness, subjects, topics and point of
view. In C. Li (Ed.), Subject and topic. New York: Academic Press.
Clark, H. H., & Carlson, T. B. (1981). Context for comprehension. In J. Long & A.
Baddeley (Eds.), Attention and performance IV. Flillsdale, NJ: Lawrence
Eribaum.
Clark, H. H., & Clark, E. (1977). Psychology and language. New York: Harcourt
Brace Jovanovich.
Clark, H. H., & Haviland, S. E. (1977). Comprehension and the given-new
contrast. In R. Freedle (Ed.), Discourse production and comprehension.
Norwood, NJ: Ablex.
Clark, H. H., & Marshall, C. R. (198 1). Definite reference and mutual knowledge.
In A. K. Joshi, B. L. Webber, & I. A. Sag (Eds.), Elements of discourse
understanding. Cambridge: Cambridge University Press.
Clark, H. H., & Schaefer, E. (1987). Collaborating on contributions to
conversations. Language and Cognitive Process, 2, 19-41.
Clark, H. H., & Schaefer, E. (1989). Contributing to discourse. Cognitive Science,
13, 259-294.
Clark, H. H., & Wilkes-Gibbs, D. (1986). Referring as a collaborative process.
Cognition, 22, 1-39.
Cutler, A. (1976). Phoneme monitoring reaction time as a function of preceding
intonation contour. Perception and Psychophysics, 20, 55-60.
Cutler, A., & Foss, D. J. (1977). On the role of sentence stress in sentence
processing. Language and Speech, 20, 1-10.
Di Cristo, A. (1995). French Intonation. In D. Hirst & A. Di Cristo (Eds.),
Intonation systems: a survey of twenty languages. Cambridge: Cambridge
University Press (in press).
Dooling, D. J., & Mullet, R. L. (1973). Locus of thematic effects in retention of
prose. Journal of Experimental Psychology, 97, 404-406.
Goffman, E. (1973). La mise en scène de la vie quotidienne (Relations in public).
Paris: Editions de Minuit.
Goffman, E. (1974). Les rites d'interaction (Interaction rituals). Paris: Editions de
Minuit.
Goffman, E. (1981). (French translation, 1987). Façons de parler (Forms of talk).
Paris: Editions de Minuit.
Halliday, M. A. K. (1967). Notes on transitivity and theme in English, Part 2.
Journal of Linguistics, 3, 199-244.
Halliday, M. A. K. (1970). Language structure and language function. In J. Lyons
(Ed.), New horizons in linguistics. New York: Penguin Books.
Haviland, S. E., & Clark, H. H. (1974). What's new? Acquiring new information as
a process in comprehension. Journal of Verbal Learning and Verbal
Behavior, 13, 512-521.
Hornby, P. A. (1971). The role of topic-comment in the recall of cleft and
pseudocleft sentences. Papers from the Seventh Regional Meeting of the
Chicago Linguistic Society (pp. 445-453).
Hornby, P. A. (1973). Intonation and syntactic structure in the development of
presupposition. Paper presented at the biennial meeting of the Society for
Research in Child Development, Philadelphia.
Hornby, P. A. (1974). Surface structure and presupposition. Journal of Verbal
Learning and Verbal Behavior, 13, 530-538.
Hornby, P. A., & 1-lass, W. (1970). Use of contrastive stress by preschool
children. Journal of Speech and Hearing Research, 19, 395-399.
Hupet, M. (1988). Syntactic and prosodic marking of pragmatic information in 4 to
10 year olds: A view from the production of contrastive information.
Second World Basque congress, Tl (397-41 0). Vitoria-Gasteiz: Servicio
central de publicaciones del Gobierno Vasco.
Hupet, M., & Tilmant, B. (1986). What are clefts good for? Some consequences
for comprehension. Journal of Memory and Language, 25, 419-430.
Knowles, J. (1986). The cleft sentence: a base-generated perspective. Lingua,
69, 295-317.
Massip, C., & Vilm, I. (1987). Comparison de I'efficacité relative de la
construction clivée et de l’accent pour contraster une information dans un
dialogue. Research note. Aix-en-Provence: Université de Provence.
Maratsos, M. (1 973). The effect of stress on the understanding of pronominal
coreference in children. Journal of Psycholinguistic Research, 2, 1-8.
Moreau, M. L. (1976). C'est: étude de syntaxe transformationnelle. Université de
Mons: Editions Universitaires de Mons.
Needham, W. P. (1990). Semantic structure, information structure and intonation
in discourse production. Journal of Memory and Language, 29, 455-468.
Pechmann, T. (1984). Accentuation and redundancy in children's and adults'
referential communication. In H. Bouma & Don G. Bouwhuis (Eds.),
Attention and performance X Control of language processes. London:
Lawrence Eribaum Publishers.
Prince, H. (1978) A comparison of WH-clefts and IT-clefts in discourse.
Language, 54, 883-906.
Rossi, M. (1985). L'intonation et l'organisation de l'énoncé. Phonetica, 42, 135153.
Somicola, R. (1988). It-clefts and Wh-clefts: two awkward sentence types.
Journal of Linguistics, 24, 343-379.
Steedman, M. J., & Johnson-Laird, P. N. (1980). The production of sentences,
utterances and speech acts: Have computers anything to say? In
Butterworth (Ed.), Language Production, Vol. I. London: Academic Press.
Terken, J., & Nooteboom, S. G. (1987). Opposite effects of accentuation and
deaccentuation on verification latencies for given and new information.
Language and Cognitive Processes, 2, 145-163.
Tilmant, B., & Hupet, M. (1990). Effects du degré de conviction de l'interlocuteur
sur l'usage de phrases clivées. L’ Année Psychologique, 90, 213-229.
Touati, P. (1987). Structures prosodiques du Suédois et du Français. Lund: Lund
University Press.
Vion, M., & Amy, G. (1984). Comprendre les relations agent-patient dans les
énoncés simples en français, une étude génétique du traitement des
structures clivées. Archives de Psychologie, 52, 209-229.
Welsh, A. (1980). The effect of word accent and syllabic stress on the processing
and perception of continuous speech; Unpublished Ph.D. thesis. Chicago:
University of Chicago.